Flatbed vs drum scan

Hi,

<p>

Would someone please tell me if flatbed scan of large format B/W photographs is as good as or even better than drum scan? If flatbed is recommended, what type? Also, what resolution do I need for scanning of 1:1 high quality reproduction? Is color scan recommended or duo/Tritone, etc.?

<p>

This is rather new to me. Your response will be helpful.

<p>

Many Thanks,

<p>

Aaron

Hi Aaron, let me start with the remark that it will be pretty hard to

find a person in Singapore who is capable of producing a decent drum

scan. I had the same problem you mention, and will stick with a 4x5

inch Pro Photo CD scan.

Having said that, a 4x5 inch drumscan of a negative or slide (as

opposed to "photograph" you mentioned) is by right "better" than a

flat bed scan. However, it all depends on the purpose of your final

product.

Do you require a tack sharp huge enlargement (remember the Louis

Vuitton banner on Ngee Ann City a few months ago), than a drum scan

is required. It will set you back around USD 100 per scan.

For anything else, up to say A1 format, a high quality flatbed scan

of your slide/negative should do. A 4x5 inch slide scanned at 1500

dpi will provide you a 20x25 inch result at 300 dpi! Another thing to

consider is D-Max which is practically the contrast range of the

scanned file. Any flatbed scanner that scans on a dynamic range of

more than 3.5 (5 is max) should do. Also note that the scanner should

be able to scan your 4x5 inch slide/neg at an adequate optical

resolution; it should not interpolate its files.

<p>

Cheers,

<p>

Mark

Although good drum scans will always beat good flatbed scans, there

was a very interesting discussion on the Piezography 3000 user group

a month or two back on how best to maximize flatbed scans. The

consensus seemed to be that by using Kami mounting fluid on a good

flatbed (such as the Epson 1680 series) with good software (like

Silverfast) and when using large format (4 x 5 or larger) one can

approach drum scan quality. You should probably find the user group

(on Yahoo groups) and search through the archives.

the principal difference between the results you get from a $50,000

drum scanner and a $350 Epson scanner are (1) sharpness and (2) shadow

detail. if you are going to all the effort to shoot 4x5's with decent

equipment, then using an Epson flatbed scanner for your printing is

like putting a vaseline-smeared plastic filter over your schneider

lens.

<p>

there are some good flatbed scanners out there (one of the best is

made by Scitex, and it costs something like $7000), but please don't

be fooled by the apparently good results you get from a $350 Epson

scanner-- they look great on your screen, but if you compare apples to

apples at 100%, you'll realize that using a cheap flatbed scanner is

no substitute.

<p>

of course, it is also important what your final results are intended

to be. if you are scanning for the web, you will get just as good

results with a cheap scanner as you would with a drum scanner. but if

you're printing, even fairly small prints, then there really is no

comparison.

I've tried the Kami on Epson 1680 technique that Howard was talking

about (although I haven't seen the discussion on Yahoo groups yet).

IMHO, it has more benefit for scanning MF material than LF. It's the

only way I've found to create an acceptably flat scan of roll film

given the natural curl of the film and the poor excuse for a MF film

holder which Epson provides with this scanner. I found it very

difficult (or sometimes impossible) to work the bubbles from the

edges of LF material. The heavy weight of the film base and the

presence of processing clip deformations at the film edges makes LF

tough to work with in some situations.

<p>

As to the scanner itself... The Epson 1680 is impressive given it's

cost, (Chris - it's not a $350 model - it's more like a $1000ish

machine) however it's not in the same league as a high-end CCD or

drum scanner. If your transparency doesn't push beyond the Dmax-Dmin

range of the scanner, the Epson 1680 can do a credible job. However,

it doesn't matter what tricks you try with mounting fluids or the

like, if the transparency has a broad range, you can't make any

scanner see more than it's capable of doing. With that said, the

coming years may be a challenge for the high-end CCD (e.g. Imacon)

scanners. The Epson 1680 is an example of a fairly well implemented

(except the film holders) lower-end scanner which is closing the

image quality gap between low-end and high-end devices much faster

than the high-end is extending their capabilities. One last note on

the 1680 - don't waste your money on the Firewire version; it doesn't

seem to make scanning any faster.

I'd be interested in knowing how a mounting fluid could be used with a

scanner like the 1680. I just don't see how this is possible. First,

the film holders do not allow the film to touch the glass surface.

Secondly, the film/transparency mode does not focus at the glass, but

above the glass at the film surface.

I totally disagree with the statement regarding the film holders. Used

properly, the holders will keep the film as flat as any enlarger

glass-less negative carrier.

As to deformations from film clips, why do you need film clips for

sheet film? I have never had a 4x5 negative curl or deform when

hang-drying. Never.

Matt,

<p>

The 1680 has two focus points - 0 mm above the flatbed and 2.5mm

above the flatbed. You can choose either focus point from within

SilverFast on their vertical icon-based toolbar.

<p>

The film clip deformations are from the holders used in a dip and

dunk processor by commercial photo labs to process E6. Unless you do

your own E6 in a Jobo type machine, you end up with these clip marks.

<p>

I'd guess one of the advantages of using a mounting oil is it's the

best way to keep the film absolutely flat when scanning and not have

newton rings. The 1680 4x5 holder is OK, but it's still flimsy

plastic and it doesn't really grip the film tight around the edges.

Given that 4x5 film is pretty rigid and flat on it's own, the end

result is OK. I've seen much better holders on other scanners (e.g.

Imacon and the Umax rubber/metal holders). The Epson holder that I

really dislike is the roll film holder. It's hard to believe that

the film flatness when using their roll film holder is the same as

you'd get out of a good enlarger. Epson designed a generic 6x17

holder which they expect you to use for 6x4.5, 6x6, 6x7, 6x8, 6x9,

6x12 and 6x17. It certainly reduced the number of holders Epson had

to provide, but the one they did provide is all but useless IMHO.

There are some points to watch for when choosing a scanner. D-Max is one, true optical resolution is another but

hardware optic quality, which has often nothing to do with resolution itself, is the key to crisp scans. Some (relative)

low cost scanners have a high bit sampling, upto 48 bits and I would say they produce astonishing color accuracy.

Their D-Max upto 3.7 and the possibility to adjust the gamma for dense images allows for very nice scans, at least on

the computer screen. The scanner I have used in a 3K price range is the Quato X-Finity 48 (PFU). This scanner is very

productive, uses SilverFast and has ICC calibration. You get a good scan from a good image. At least that is what an

untrained eye believes. But when you compare with a high-end flatbed such as an Eversmart, a scanner that

produces drum scan quality, you soon realize where some of the differences that (partly) justifies the much higher

price tag lay. First, an antiblooming CCD for contrasty and noisefree scans, second an excellent sharp optic for crisp

details right into the film grain, third a D-Max of 4.1 for detailed shadows and fourth an engine that privilegiates

quality over speed.

<p>

Now you take the two scans and put them side by side. On the screen, at small size, all you can see is that the high

end image is a little more brilliant and has richer tonal values in the shadows. Then you enlarge them and the

sharpness difference appears. Even if the scan resolution was the same, one looks like a large format image when

the other seem to come from a small or medium format scan in comparison. At this point, you try to apply unsharp

masking to the low end scan to match the high end scan, but noise appears. Lots of clear pixels in the shadows.

<p>

So, yes, a low end flatbed can be as good as a drum scan for presenting an image on the web or even for a small

print. And a high end flatbed (Fuji Lanovia, Eversmart Supreme) can be as good as the best drum scanner for a high

resolution scan. And hardware is just one aspect of the final scan quality. Much depends of the skills of the operator

too! I understand, Aaron, that you want to scan original B&W prints for offset reproduction? Then a good flatbed will

probably do the best, there is no need to go for a very high quality scanner, as prints have not the depth of a

negative or slide anyway. RGB, CMYK, this depends on the process used for the print. See your printer for details. If

it's to print on an inkjet printer, an RGB scan will probably be a good start. Some inkjet printers make much nicer

B&W prints from RGB colors than from black ink only, and otherwise, you can always convert the RGB scan into gray

in Photoshop. Get as much resolution as you can for the price, but a minimum of 350 dpi at 1:1. It seems that beyond

700 dpi the difference won't be too noticable on most printers. For offset, a factor quality of 200% of the print

screen is plenty (for example 350 dpi for the highest quality raster of 175 lpi ) A last word on unsharp masking: If the

scan is for multiple uses and sizes, ask a for an unsharpened scan and apply the filter in Photoshop when the size of

the output has been defined. Using a pixel ray of 0.3 and a theshold of 0 with an ammount of 200 to 500 works well

for most scans.

This whole discussion does not make sense. I have an HP Photosmart

that I modified to accept 4x5 negatives and transparencies. (Not

hard) The film transports easily into the machine and the scans are

incredible. They are as good as any drum scan I have ever seen.

<p>

The hype over drum scanning was invented by the manufacturers to

separate you and your money!

<p>

I would put one of my Photosmart scans up angainst any drum scan, any

day, for any amount of money!

<p>

Adam.

I would like to thank all for their kind response. I'm absorbing the

very useful information although I need to look up the meaning of

some terms used here. Appreciation also to Paul Schilliger for his

detailed and most direct answers to my questions. You've all been

very helpful.

<p>

Aaron

I would be interested in knowing how Adam modified the HP Photosmart

to accept 4x5 tranies.

<p>

Thanks

Yeah, I'd like to know how a Photosmart can be converted to scan 4x5

as well. I heard of folks slicing up their film into strips, but

that's out of the question for me.

I'ld like to know if the Epson 1680 can be succesfully used for

scanning 6x17 color and b/w transparencies, color negs and b/w negs?

<p>

Anyone done that successfully? I wan tot make pritn son my Epson

3000, size: about 44 inches maximum length, and about 17 inches max

height.

<p>

THANKS

I would be interested in knowing how Adam modified the HP Photosmart

to accept 4x5 tranies.

----------------------------------------------------------------------

Yeah, I'd like to know how a Photosmart can be converted to scan 4x5

as well. I heard of folks slicing up their film into strips, but

that's out of the question for me.

<p>

Me, too. I have a Microtek Scanmaker 4 that I use for anything bigger

than 35mm, but I've got an old Photosmart that still gives better

35mm scans than the flatbed (duh). Using it for bigger films sounds

like a great idea.

<p>

Tony

Sorry Adam, I don't think you've seen the results of really good drum scans.

<p>

Consider the specs: HP Photosmart has 2400dpi optical resolution. DMax

(maximum density) is unspecified.

<p>

Drum scanners typically start at 4000dpi - and up to 25,000 for a serious one.

DMax is up to 4.2.

<p>

The photomultiplier tubes used in drum scanners simply have a much wider

dynamic range than the very best CCDs. This does make a difference that will

be visible in many instances.

<p>

That said, for the original questioner's purpose, the difference wouldn't be

that great. But the 2400dpi resolution of the HP limits enlargements to about

8X with a resolution of 300dpi � so a 35mm image would be roughly the size

of a letter.

<p>

I'd be the first to agree a drum scan is overkill for many applications, but to

say it can be matched by any CCD scanner is simply technically incorrect.

Apart from the machines' capabilities, an experienced professional drum

scanner operator will be able to get better results out of any scanner than

the most conscientious part-timer.

<p>

And I'm surprised to hear these are hard to find in Singapore, given the

volume of high quality printing done there.

<p>

As for the price of drum scanners - it's simply driven by volume.

<p>

There's no way I'd suggest everyone goes for a drum scanner, or that their

quality is necessary for every application. But for critical print reproduction

work or significant enlargements, they're still essential.

<p>

I don't know what it's like elsewhere, but here in Sydney drum scans are

cheaper than ever. If you're doing a few with reasonable deadlines you can

get scans done for A$20-30 (US$10-15) each. Very reasonable when you

consider you're getting a professional operator and a machine worth

A$100,000+.

<p>

Hope this is useful. (Incidentally, I'm not in the scanning business, but have

been buying high quality printing for many years).